Lack of testicular seipin causes teratozoospermiasyndrome in menMin Jianga,1, Mingming Gaob,1, Chaoming Wuc,1, Hui Hea, Xuejiang Guoa, Zuomin Zhoua, Hongyuan Yangd,Xinhua Xiaoe,2, George Liub,2, and Jiahao Shaa,2

aState Key Laboratory of Reproductive Medicine, Department of Histology and Embryology, Nanjing Medical University, Nanjing 210029, China; bInstitute ofCardiovascular Sciences and Key Laboratory of Molecular Cardiovascular Sciences, Ministry of Education, Peking University Health Science Center, Beijing100083, China; cDepartment of Endocrinology, The Second Affiliated Hospital of Wenzhou Medical University, Wenzhou 325027, China; dSchool ofBiotechnology and Biomolecular Sciences, University of New South Wales, Sydney, NSW 2052, Australia; and eDepartment of Endocrinology, Key Laboratoryof Endocrinology, Ministry of Health, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College,Beijing 100730, China

Edited by John J. Eppig, The Jackson Laboratory, Bar Harbor, ME, and approved April 1, 2014 (received for review January 2, 2014)

Obesity impairs male fertility, providing evidence for a link be-tween adipose tissue and reproductive function; however, poten-tial consequences of adipose tissue paucity on fertility remainunknown. Lack of s.c. fat is a hallmark of Berardinelli–Seip congen-ital lipodystrophy type 2 (BSCL2), which is caused by mutationsin BSCL2-encoding seipin. Mice with a targeted deletion of mu-rine seipin model BSCL2 with severe lipodystrophy, insulin re-sistance, and fatty liver but also exhibit male sterility. Here, wereport teratozoospermia syndrome in a lipodystrophic patientwith compound BSCL2 mutations, with sperm defects resemblingthe defects of infertile seipin null mutant mice. Analysis of condi-tional mouse mutants revealed that adipocyte-specific loss of sei-pin causes progressive lipodystrophy without affecting fertility,whereas loss of seipin in germ cells results in complete male in-fertility and teratozoospermia. Spermatids of the human patientand mice devoid of seipin in germ cells are morphologically abnor-mal with large ectopic lipid droplets and aggregate in dysfunc-tional clusters. Elevated levels of phosphatidic acid accompaniedwith an altered ratio of polyunsaturated to monounsaturated andsaturated fatty acids in mutant mouse testes indicate impairedphospholipid homeostasis during spermiogenesis. We concludethat testicular but not adipose tissue-derived seipin is essentialfor male fertility by modulating testicular phospholipid homeostasis.

Spermatogenesis encompasses many complex processes, suchas the proliferation of spermatogonia, meiosis of spermato-

cytes, and morphological changes associated with spermatidformation. These processes are modulated by factors that areboth intrinsic and extrinsic to the testis. Intrinsic factors derivefrom genes predominantly expressed in testis, including DEAD(Asp-Glu-Ala-Asp) box polypeptide 4 (DDX4), synaptonemalcomplex protein 3, and deleted in azoospermia 1 among others,whereas extrinsic factors originate from the hypothalamus, pi-tuitary gland, and other organs. Recent studies have shown thatthe secretion of adipokines, such as leptin from adipose tissue,regulates reproductive function by affecting the hypothalamus–pituitary–gonadal (HPG) axis (1, 2). Evidence for a link betweenadipose tissue function and male fertility has also emerged fromseveral studies showing a negative impact of obesity on malereproduction (3, 4). Whether the opposing syndrome, loss of adi-pose tissue, or lipodystrophy impacts on male infertility remainslargely unexplored. A single metaanalysis evaluating body massindex and male semen quality reported that underweight was as-sociated with an increased but nonsignificant risk of abnormalsperm count (5). A study on several types of lipoatrophies men-tioned that men with congenital generalized lipodystrophy (CGL)were fertile, but it did not present specific data (6). CGL, alsoknown as Berardinelli–Seip congenital lipodystrophy (BSCL), isa rare autosomal recessive disease characterized by near-completeatrophy of general adipose tissue and hypertriglyceridemia afterbirth as well as hyperinsulinemia and hepatomegaly becauseof hepatic steatosis. Mutations in two genes, 1-acylglycerol-3-

phosphate O-acyltransferase 2 and BSCL2, are causative in CGLtypes 1 and 2, respectively; 1-acylglycerol-3-phosphate O-acyl-transferase 2, localized on human chromosome 9q34, encodes anenzyme participating in biosynthesis of triglyceride and glycer-ophospholipids (7, 8), whereas BSCL2 on chromosome 11q13encodes seipin, a transmembrane protein of unknown functionthat localizes to the endoplasmic reticulum of adipocytes andother cells (9, 10). Seipin is highly expressed in human brain,testis, and adipose tissue (9). Lack of seipin in the brain has beenlinked to motor neuropathy and Silver syndrome (11–13), andloss in adipose tissue has been linked with severe lipodystrophy(14); its role in testis remains unknown.In our previous study, we made the surprising observation that

seipin-deficient mice not only model BSCL2 with marked lip-odystrophy (15) but also, exhibit complete male infertility. Toevaluate a link between lipodystrophy and male fertility andassess the particular role of seipin, we examined reproductivefunction and sperm quality in male CGL type 2/BSCL2 patientsand characterized mice with adipose- and germ cell-specific lackof seipin. We report that seipin deficiency causes teratozoospermiain humans and mice. Specifically, testicular but not adipose tissue-derived seipin is required for male fertility, playing a central rolein modulating the testicular phospholipid (PL) homeostasis re-quired for normal spermiogenesis.

Significance

The relationship between body fat and male reproduction isclearly seen when excess fat compromises fertility; however,potential consequences of adipose tissue paucity on fertilityare unclear. We report that lack of seipin, a transmembraneprotein localizing to the endoplasmic reticulum, causes bothpaucity of adipose tissue and male sterility. Human patientsand mouse models lacking seipin in germ cells produce severelyabnormal sperm because of impaired lipid distribution duringsperm maturation. These defects are testis-intrinsic; scarcityof adipose tissue per se does not affect male fertility. Ourwork shows that two distinct organs—adipose tissue andthe male germ line—depend on seipin function and thatappropriate lipid distribution is essential for production ofmature, functional sperm.

Author contributions: M.J., Z.Z., H.Y., X.X., G.L., and J.S. designed research; M.J., M.G.,C.W., and H.H. performed research; M.J., M.G., C.W., X.G., Z.Z., X.X., G.L., and J.S. analyzeddata; and M.J., G.L., and J.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1M.J., M.G., and C.W. contributed equally to this work.2To whom correspondence may be addressed. E-mail: shajh@njmu.edu.cn, vangeorgeliu@gmail.com, or xiaoxinhua@medmail.com.cn.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1324025111/-/DCSupplemental.

7054–7059 | PNAS | May 13, 2014 | vol. 111 | no. 19 www.pnas.org/cgi/doi/10.1073/pnas.1324025111

Dow

nloa

ded

by g

uest

on

Mar

ch 3

1, 2

020

ResultsSevere Teratozoospermia Syndrome in a Male Patient with CompoundSeipin Mutations. To investigate the relationship between lipodys-trophy and male fertility, we studied a male patient from a familyaffected by lipodystrophy. The patient and his older sister exhibitedlipodystrophy; mother, stepfather, and stepsister were not affected.Sequencing of all exons and exon/intron boundaries of BSCL2revealed compound heterozygous mutations in the affected patients(Fig. 1 A and B), a 19-bp deletion from nucleotide 358 to 376in exon 2, and a G to T mutation at nucleotide 757 (codon 253)in exon 5; these results were further confirmed by sequencing ofsubcloned PCR products (Fig. 1 C and D). The 19-bp deletion,inherited from the heterozygous mother, caused a frameshift(S119fsX155) and thus, premature termination of seipin, and thepaternally inherited mutation of codon 253 replaced glutamic acidwith a stop codon (E253X) (Fig. 1 B and C).On admission, the patient presented with an extremely muscular

and acromegaloid appearance; umbilical hernia and acanthosisnigricans over the axillae were also noted (Fig. 1E). A physicalexamination revealed enlarged penis and a testicular volume of32 mL bilaterally. The patient had high fasting blood glucoselevels (21 mM/L). Sex hormones were in the normal range,albeit at the lower end for FSH and luteinizing hormone (LH).

The patient was diagnosed with teratozoospermia syndrome bytwo separate reproductive centers according to World HealthOrganization reference values for human semen characteristics(16) (Table 1). Sperm defects included abnormal head mor-phology and the presence of bundled sperm with two or moresperm connected to each other (Fig. 2A). Oil Red O staining ofsperm samples revealed that heads of bundled sperm containedlarge ectopic lipid droplets (LDs), which are not present innormal sperm (Fig. 2B).

Seipin Antibody Generation and Stage-Specific Expression of Seipin inthe Testis. To characterize the distribution of seipin in testiculartissue, we used a rabbit polyclonal antibody generated againstthe C-terminal domain of seipin. The specificity of this antibodywas confirmed by Western blot analysis detecting specific bandsin extracts from testis and adipose tissue from WT littermatemice (S-WT) but not mice with ubiquitous deletion of seipin(S-KO) (Fig. 3A). In extracts from human testis tissue, the an-tibody detected two bands at 55 and 70 kDa in accordance withprevious analyses of the two known human isoforms that resultfrom both alternative splicing and alternative translation initia-tion start site use on human transcripts (17) but absence of re-activity in isolates from mature sperm (Fig. 3A). In the mouse,a long (NM_001136064.2) and a short (NM_008144.4) transcriptof seipin have been identified; they result from differential splicingof a 5′ UTR exon. By RT-PCR analysis, we detected higher ex-pression levels and marked predominance of the long transcriptin murine testis tissue, whereas murine adipose tissue containedlow levels of both transcripts (Fig. S1). Because both transcriptsare predicted to encode the same protein (443 aa), the observedsize difference of seipin protein from murine testis (between 55 and70 kDa) and adipose tissue (55 kDa) (Fig. 3A) may be becauseof different posttranslational modifications in different tissues.Immunohistochemical analysis of human and murine testicu-

lar tissue confirmed stage-specific expression of seipin duringspermatogenesis. In humans, seipin was present in the cytoplasmof late spermatocytes to spermatids but absent from maturesperm (Fig. 3B), whereas murine seipin appeared at later phasesof spermatogenesis and was detectable in steps 7–16 spermatidsbefore incorporation into the residual body (Fig. 3C). Humanand mouse seipin transcript and protein were also detected inLeydig cells at a lower level compared with germ cells (Fig. S2).

Loss of Testicular but Not Adipose Seipin Causes Male Infertility. Toinvestigate the consequences of tissue-specific loss of seipin onmale reproduction, we crossed mice with a floxed seipin allele(exon 3;E3fl/fl) (15) with mice expressing Cre recombinase ingerm cells (DDX4-Cre mice) and adipocytes (adipocyte protein2-Cre mice), generating mice with germ cell-specific (gS-KO)and adipocyte-specific (aS-KO) deletion of seipin, respectively.S-KO mice were obtained by intercrossing seipin+/− mice (15).S-KO males displayed severe lipodystrophy and absence of seipinprotein in both somatic and germ cells of the testis, whereasgS-KO males had normal levels of adipose tissue and lackeddetectable seipin in germ cells but not Leydig cells of the testis

Fig. 1. BSCL2 mutations in a family with congenital lipodystrophy. (A)Family tree of male patient 1 (P1) and five family members. P1 and his sister[patient 2 (P2)] exhibited lipoatrophy; his mother, stepfather, and stepsisterwere not affected. P1 and P2 were compound heterozygotes for a G to Tmutation in exon 5 (green) and a 19-nt deletion in exon 2 (red) of the BSCL2gene. The mother was heterozygous for this 19-nt deletion. (B) Detection ofmutations by sequence analysis of the BSCL2 gene. A heterozygous 19-bpdeletion from nucleotides 358 to 376 in exon 2 in both the subject and herbrother causes a frameshift (S119fsX155). The G to T mutation in exon5 converts codon 253 (glutamic acid) to a stop codon (E253X), resulting inpremature termination. Ctl, control sample from an unrelated healthy man; M,mother. Arrowheads indicate the first base of the respective mutation.(C and D) The proportion of mutant to normal sequence in patient samplesindicates heterozygosity for each mutation. PCR fragments amplified frompatient DNA were subcloned followed by sequencing of plasmids isolated inrandomly picked colonies. (E) Physical appearance of P1 with marked ab-sence of s.c. fat and muscular appearance.

Table 1. Hormone levels and sperm abnormalities in a male lipodystrophic patient with BSCL2mutations

HormoneConcentration

(reference value*)Sperm morphology

(200 sperm) Center I (%) Center II (%)

LH 2.92 (1.24–8.62) IU/L Normal (%) 2.50 2.0FSH 2.96 (1.27–19.46) IU/L Head defect (%) 95.50 95.0Estradiol 15.00 (0–50) pg/mL Neck defect (%) 53.00 43.0Testosterone 4.94 (1.75–7.81) ng/mL Tail defect (%) 22.50 15.0

*Reference values used by the laboratory of the Second Affiliated Hospital of Wenzhou Medical University.

Jiang et al. PNAS | May 13, 2014 | vol. 111 | no. 19 | 7055

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Mar

ch 3

1, 2

020

(Fig. 4 A–C). The aS-KO males showed progressive lipodystrophy(14) but normal levels of seipin in testis (Fig. 4 A–C).Males with the three different seipin mutations (S-KO, gS-KO,

and aS-KO) exhibited normal mating behavior, evident fromcopulatory plugs in WT females after overnight mating. How-ever, during a 3-month fertility test, no pregnancies or offspringwere observed in breeding pairs of WT females and S-KO orgS-KO males, whereas aS-KO males sired offspring at approxi-mately the same frequency as WT controls (Table S1).Serum levels of sex hormones, including FSH, LH, testoster-

one, and estradiol, were normal in S-KO, gS-KO, and aS-KOmales (Fig. S3). However, both S-KO and gS-KO mice had sig-nificantly lower sperm counts and percentages of motile spermcompared with littermate controls (Table S2). Sperm from S-KOand gS-KO males was morphologically highly abnormal, withhead and tail abnormalities in 60% ± 15% and 18% ± 6% ofsperm, respectively, whereas head or tail defects were found inonly 6% ± 4% of sperm from respective littermate control mice(Fig. 4D). Defects in sperm heads included lack of the usualhook and banana-like or amorphous forms (Fig. 4 E, b and c).

Tail defects manifested with folded sperm or as sheath abnor-malities (Fig. 4 E, b and e). Some sperm exhibited cavities in thesperm head (Fig. 4 E, d). Similar to the teratozoospermia syn-drome of patient 1, testes from S-KO and gS-KO males con-tained prominent bundles of two or more interconnected sperm(Fig. 4 E, f). In contrast, aS-KO males had normal sperm, similarto normal littermate controls (Fig. 4D and Table S2). Normalmale reproductive function in aS-KO males but infertility ofS-KO and gS-KO males indicates that loss of seipin functionin germ cells causes male infertility, specifically by causing oli-goasthenoteratozoospermia. To verify that these observationswere not related to the metabolic status of the males, we assessedfasting plasma levels of serum cholesterol (total cholesterol),triglycerides, and glucose and serum levels of insulin at fed state.None of these metabolic markers differed significantly betweengS-KO and control mice (Fig. S4). S-KO and aS-KO mice hadhigher total cholesterol levels and significantly elevated fedplasma insulin levels compared with controls, whereas fastingtriglycerides and glucose levels of all mutants were similar tocontrols (Fig. S4). The absence of detectable metabolic ab-normalities in gS-KO mice further confirms that the infertilityof seipin-deficient mice is not caused by metabolic changesbut because of the absence of seipin protein in germ cells.

Lack of Seipin in Germ Cells Causes Structural Defects in LateSpermiogenesis. In the three seipin-deficient mouse models, sper-matogonia and spermatocytes did not exhibit obvious morpho-logical differences. However, in S-KO and gS-KO males, weobserved abnormal clustering of late spermatids of seminiferoustubules at stages VII and VIII of the seminiferous epithelial cycle(Fig. 5A). The heads of these spermatids were randomly ori-ented, whereas the heads of the spermatids from the WT controlswere aligned perpendicular to the seminiferous tubule basementmembrane. S-KO and gS-KO tubule sections near completion ofspermiogenesis contained massive accumulations of spermatidsin bundle-like structures, suggestive of defects during spermatidindividualization, which defines the final stage of normal sper-miogenesis. The numbers of spermatocytes in stages VII and

Fig. 2. Teratozoospermia phenotype of P1. (A) Abnormal sperm heads(arrowheads), bundled sperms (arrows), and bundled tails (long arrow). (B)Ectopic accumulation of LDs in sperm of P1. Blue, nuclei; red, LDs.

Fig. 3. Stage-specific expression of seipin in humanand murine testis. (A) Western blot analysis usinga rabbit polyclonal antibody raised against theC-terminal end of seipin; this antibody detects aspecific seipin band in testis (between 55 and 70 kDa)and adipose tissue (at 55 kDa) from S-WT (arrows) butnot from S-KO mice. In human testis tissue, the anti-body detects two bands at ∼55 and 70 kDa, consistentwith the two predicted human isoforms, whereasmature human sperm lacks seipin reactivity. Theloading control for murine samples is β-actin, andfor human samples, it is β-tubulin. HS, human sperm;HT, human testis. (B) In human testis, seipin is presentin later spermatocytes (Sc) and round (rSt) and elon-gating (eSt) spermatid stages but not spermatogonia(Sg) or mature sperm (S). (C) During murine sper-matogenesis, seipin is detectable in steps 7–16 sper-matids during spermiogenesis.

7056 | www.pnas.org/cgi/doi/10.1073/pnas.1324025111 Jiang et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 3

1, 2

020

VIII seminiferous tubules from S-KO, gS-KO, and aS-KO micewere similar to the numbers in S-WT mice, but tubules fromS-KO and gS-KO mice contained significantly fewer round sper-matids (Fig. S5A). We speculate that the loss of round spermatidswas caused by disruption of the release of step 16 spermatids thatare normally not bundled and become released into the lumen ofthe seminiferous tubule. Because of the bundle-like aggregation ofmost of these spermatids in the mutant, disengagement wouldlikely be aberrant and may cause defects in the architecture of theseminiferous tubule wall. Cauda epididymides from S-KO andgS-KO mice were almost entirely devoid of sperm but containednumerous spherical cells, whereas aS-KO males and WT controlshad similar numbers of epidydimal sperm (Fig. 5A). The mor-phological abnormalities associated with lack of seipin in germcells, therefore, originate from disturbances in morphologicalchanges associated with spermiogenesis.

Lipid Accumulation in Sperm Caused by Altered PL Metabolism inSeipin-Deficient Testis. Similar to the ectopic lipid spheres ob-served in sperm from patient 1, testicular tissue from S-KO andgS-KO mice contained large ectopic LDs in the perinuclear re-gion of round to elongated spermatids. These LDs were presentin every tubule of each section assessed and predominated inelongating spermatids, which exhibited the highest expressionlevels of seipin during spermatogenesis (Fig. 5B and Fig. S5B). Incontrast, only small LDs colocalized with residual bodies ofelongated sperm in WT and aS-KO testes (Fig. 5B). Electronmicroscopic analyses of S-KO testes confirmed the presence oflarge ectopic LDs in round, elongating, and elongated sper-matids (Fig. 6 A and B). A similar phenotype was observed ingS-KO mice (Fig. S5C), whereas in WT and aS-KO testes, onlyLDs of small size were present, located within residual bodies.Spermatids from S-KO and gS-KO mice exhibited defects inchromatin condensation (Fig. 6B), and S-KO and gS-KO testescontained many degenerating masses composed of fragmented

spermatids and large LDs (Fig. 6C). Clusters of spermatid tailswere present and appeared as bundles in longitudinal sections(Fig. 6C). Purification of haploid cells with ectopic large LDs foradditional analysis was not feasible, and Oil Red O staining ofLDs in Leydig cells did not reveal any apparent differences be-tween testis tissue from S-WT, S-KO, gS-KO, and aS-KO mice(Fig. S6). We, therefore, performed lipidomic analysis of PLs byliquid chromatography–tandem MS of whole testes from S-KO,gS-KO, and S-WT males. Analyses of fatty acid compositionsand their relative levels in testes tissue were assessed for glycero-PLs (GPL), including phosphatidic acid (PA), phosphatidylcholine(PC), phosphatidylethanolamine, phosphatidylserine, phosphati-dylinositol, phosphatidylglycerol, and total PLs. The relativecontent of PA was significantly increased in the testes of bothS-KO and gS-KO males compared with WT (133% and 46%higher, respectively; P < 0.01 and P < 0.05, respectively). PC andPLs levels were elevated in S-KO testis compared with control(19% and 16% higher, respectively; P < 0.01 compared withWT) but not significantly different between gS-KO and WT (Fig.7A). The ratios of PA to PC and PLs in both S-KO and gS-KOtestes were higher than in WT, confirming the elevated PA inS-KO and gS-KO testes (Fig. 7B). In addition, the ratio of poly-unsaturated fatty acids to monounsaturated and saturated fattyacids in different PL fractions was increased in phosphatidyl-glycerol and PL in both S-KO and gS-KO testes. For phospha-tidylethanolamine and phosphatidylserine, this ratio was reducedin both S-KO and gS-KO testes. S-KO but not gS-KO testes alsoexhibited an increased PA and decreased PC ratio, and lastly,gS-KO but not S-KO had elevated phosphatidylinositol levels(Fig. 7C). These results show elevated levels of PA and alteredPL homeostasis in S-KO and gS-KO testes, concomitant with theappearance of large ectopic LDs in sperm.

Fig. 4. Consequences of tissue-specific lack of seipin in mouse male re-productive organs. (A) Distribution of adipose tissue in mouse models withubiquitous-, germ cell-, or adipocyte-specific loss of seipin. Absence of theepididymal fat pad (arrowheads) near the testes (arrows) in S-KO and aS-KObut not gS-KO mice. (B and C) S-WT and aS-KO males express comparablelevels of seipin in testis, whereas S-KO testis and seminiferous tubules ofgS-KO males lack detectable seipin. (D) Percentage of sperm with abnormalmorphology in controls and mice with tissue-specific deletion of seipin.**P ≤ 0.01; ***P ≤ 0.001. (E ) Representative examples of abnormalmorphology of sperm from S-KO and gS-KO males. (a) Normal sperm. (b–f)Head defects include lack of the (b) usual hook, (c) banana shape, or (d )apparent cavities. Tail defects include (b) folded sperm and (e) sheathdefects (e). (f) Bundle-like masses of connected sperm. Arrows indicatemorphological abnormalities.

Fig. 5. Defective spermiogenesis in adult S-KO and gS-KO mice. (A) Tubulesfrom S-KO and gS-KO males at stages VI and VII of the seminiferous epi-thelial cycle contain clusters of late spermatids (short arrows) and bundle-likestructures (circles) without significant residual bodies. In S-WT and aS-KO tubules,spermatid heads are oriented perpendicular to the lining membrane of thetubules (arrowheads) and contain many small residual bodies (long arrows).Cauda epididymides from S-KO and gS-KO lack sperm but contain nu-merous spherical cells and vacuoles (arrows), whereas S-WT and aS-KO arefilled with sperm (arrowheads). Sections were stained with H&E. (B) OilRed O staining of S-KO and gS-KO testes reveals multiple large ectopic LDsnear the nuclei of spermatids. In contrast, S-WT and aS-KO contain smallLDs in residual bodies of the seminiferous epithelium. Arrows indicate LDs.Blue, nucleus; red, LDs.

Jiang et al. PNAS | May 13, 2014 | vol. 111 | no. 19 | 7057

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Mar

ch 3

1, 2

020

DiscussionIn the present study, we report severe teratozoospermia syn-drome in a patient with complex BSCL2 mutations predicted tocause a lack of seipin. We find that mice with ubiquitous or germcell-specific lack of seipin exhibit completely male infertility ac-companied by the same morphological abnormalities observed inthe patient’s sperm, particularly defective sperm head, ectopicaccumulation of large LDs, and aggregation in dysfunctionalbundles. These data indicate that seipin plays an important rolefor spermatogenesis and may be critical for appropriate lipiddistribution during the morphological changes associated withspermiogenesis and release of mature individual sperm.Normal spermatogenesis and fertility of aS-KO males reveal that

loss of adipose tissue because of the lack of seipin does not affectmale reproduction. Furthermore, the phenotypic differences ofaS-KO and gS-KO mice show that seipin plays separate and distinctroles in adipose tissue homeostasis and during spermatogenesis.Previous studies have shown that adipose tissue can affect

male fertility by secretion of leptin, which regulates variouscompounds of the HPG axis. Enhanced leptin secretion associ-ated with obesity presumably disturbs the HPG axis, negativelyaffecting male reproductive capacity. We find that adipocyte-specific seipin deficiency and the ensuing paucity of adiposetissue do not affect male fertility. Consistent with this observa-tion, male fertility is preserved in several other mouse models ofsystemic lipodystrophy, including mice with overexpression ofSREBP-1c in adipose tissue that have markedly reduced fatdepots (18), transgenic A-ZIP/F-1 mice lacking white fat tissue(19), or PPAR-hypomorphic mice with severe lipodystrophy andhyperlipidemia (20). Therefore, paucity of adipose tissue, despitebeing associated with low leptin levels, does not necessarily im-pair normal spermatogenesis and male fertility. Male mice withthe fatty liver dystrophy gene mutation represent an exceptionto this observation, because they exhibit both lipodystrophy andinfertility (21). However, lipin, the gene product of fatty liverdystrophy, is highly expressed in testis; therefore, male infertilitymay be caused by the loss of lipin function in testis rather thanadipose tissue dysfunction resulting in lipodystrophy (22).The hormone-sensitive lipase (HSL) gene encodes a protein

essential for male fertility; male mice lacking HSL are infertile,with testicular abnormalities resembling the abnormalities ofseipin-deficient mice, including multinuclear masses and ectopicLDs (23, 24). Defects associated with the loss of seipin or HSLemphasize a central role of appropriate lipid metabolism anddistribution during spermiogenesis. Membrane-structured organ-elles contribute to the morphological changes occurring duringspermiogenesis. The endoplasmic reticulum and Golgi apparatusfacilitate acrosome formation, and nuclear condensation andelongation of the flagellum require changes in membrane struc-ture. With completion of spermiogenesis, redundant organelles

and materials are packed into residual bodies, a process involvingthe formation of small LDs inside residual bodies and associatedwith a reduction of the volume of spermatids to ∼25% beforespermiation (25). Residual bodies become phagocytosed by Sertolicells to undergo eventual recycling, and mature spermatids sepa-rate and become released into the lumen of seminiferous tubule.

Fig. 6. Lipid abnormalities and chromatin defects inseipin-deficient male germ cells. (A) Electron micro-scopic analysis reveals large LDs in spermatids from theseminiferous epithelium of S-KO males and muchsmaller LDs in residual bodies in tubules from S-WTmales. Black dotted line outlines the contour of LDs;arrows show residual bodies. (B) In contrast to sper-matids from control littermates, round, elongating,and elongated spermatids from S-KO males containlarge ectopic LDs and exhibit defective compactionand condensation of nuclear chromatin and abnormalmorphology. (C) The seminiferous epithelium of S-KOmales contains degenerating clusters of several sper-matids and large LDs. Clusters of spermatid tails arevisualized as bundles in longitudinal sections. Arrows,LDs; arrowheads, nuclei.

Fig. 7. Altered PL homeostasis in S-KO and gS-KO testes. (A) Relative PLs level ingS-KO and S-KO testes compared with S-WT. PA was significantly increased inboth S-KO and gS-KO; PC and PLs in S-KO were also increased slightly. (B) Theratios of PA to PC and PLs were elevated in S-KO and gS-KO compared withS-WT. (C) S-KO and gS-KO testes exhibit altered ratios of polyunsaturated fattyacid (PUFA) to monounsaturated and saturated fatty acids in different PL frac-tions. PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidy-linositol; PS, phosphatidylserine. *P < 0.05; **P < 0.01; ***P < 0.001.

7058 | www.pnas.org/cgi/doi/10.1073/pnas.1324025111 Jiang et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 3

1, 2

020

The coordinated execution of such specialized events requires fineregulation of germ cell lipid metabolism, especially membranePLs, during spermiogenesis. Oresti et al. (26) showed, that withprogressing differentiation, the amount of polyunsaturatedfatty acid-rich GPL per spermatid decreased with the reduction ofcell size, whereas the ratio of fatty acids (22:5/20:4) in GPL wasincreased, but the precise requirements for particular changesin PL composition during spermiogenesis are not yet well-understood. The severe morphological defects of seipin-deficientsperm reported here likely result from disturbances in lipidmetabolism during spermiogenesis. Morphological abnormalitiesof spermatids with ectopic large LDs may be caused solely by thespace-occupying effect of such large LDs during condensation.The disturbance of PL homeostasis in spermatids could affect thestability of intercellular bridges and hence, impair the separationof matured spermatids, leading to massive accumulation of S-KOand gS-KO spermatids in bundle-like structures. Previous studieshave shown that elevated PA enhances the formation of verylarge LDs (27). PA is a cone-shaped lipid that alters the mem-brane curvature and can promote membrane fusion events. Inour lipidomics analysis, we found higher total PA and higherratios of PA to PC and PA to PL in S-KO and gS-KO testescompared with controls. Higher PA levels in S-KO comparedwith gS-KO testes may be associated with a diluting effect ofnormal lipids present in the Leydig cells of gS-KO mice, whichexpress low levels of seipin. Elevated PA levels have also beenreported from seipin-deleted yeast and seipin-KO Drosophila(27–30). We speculate that seipin may participate in the regu-lation of PA metabolism in testis, thereby controlling LD mor-phology and formation. In summary, we find that testicular butnot adipose tissue-derived seipin is essential for male fertility bymodulating testicular PL homeostasis.

Materials and MethodsThis study was approved by the Institutional Ethical Committee of NanjingMedical University (2013-26); written informed consent was obtained fromall human subjects. Exons and exon/intron boundaries of seipin (cDNANM_001122955) were amplified by PCR (Table S3 shows oligonucleotidesequences) and analyzed by direct sequencing and sequencing of fragmentssubcloned into T vector (code no. 3271; Takara)

All procedures involving mice were approved by the Animal Care Com-mittee, Peking University Health Science Center (LA2010-059). The gS-KO andaS-KO mouse strains were obtained by crossing seipin-LoxP/LoxP (15) withDDX4-Cre (006954; Jax) and adipocyte protein 2-Cre mice (14); genotypingwas performed by PCR using the oligonucleotide primers listed in Table S3.The murine plasma levels of LH, FSH, estradiol, and testosterone wereassayed at the Beijing Sino–United Kingdom Institute of Biological Tech-nology. Neutral lipid dye Oil Red O staining, histological analysis, andcomputer-assisted semen analysis were performed as described previously(31, 32) with minor modifications (SI Materials and Methods). Western blotand immunohistochemical analyses were performed using a rabbit poly-clonal antibody raised against the C terminus of murine seipin (VazymeBiotech) (SI Materials and Methods). Lipid extracts from mouse testis wereprepared and analyzed using a TSQ Vantage triple-quadrupole mass spec-trometer (Thermo Fisher Scientific) equipped with Xcalibur system softwareas described previously (33, 34). All data are presented as means ± SEMs.Statistical comparison between the two groups was performed using Stu-dent t test or one-way ANOVA. Values of P < 0.05 were consideredstatistically significant.

ACKNOWLEDGMENTS. We thank Dr. Huiyong Yin and Tao Meng forlipidomics technical assistance and Dr. Sigrid Eckardt for revising themanuscript. This work was funded by National Key Basic Research ProgramGrant 2011CB944304, Major National Basic Research Program of thePeople’s Republic of China Grants 2011CB503900 and 2012CB517505, andNational Natural Science Foundation of the People’s Republic of ChinaGrants 81222006, 30930037, and 81121061.

1. Landry D, Cloutier F, Martin LJ (2013) Implications of leptin in neuroendocrine reg-ulation of male reproduction. Reprod Biol 13(1):1–14.

2. Hamm ML, Bhat GK, Thompson WE, Mann DR (2004) Folliculogenesis is impaired andgranulosa cell apoptosis is increased in leptin-deficient mice. Biol Reprod 71(1):66–72.

3. Teerds KJ, de Rooij DG, Keijer J (2011) Functional relationship between obesity andmalereproduction: From humans to animal models. Hum Reprod Update 17(5):667–683.

4. Sermondade N, Faure C, Fezeu L, Lévy R, Czernichow S; Obesity-Fertility CollaborativeGroup (2012) Obesity and increased risk for oligozoospermia and azoospermia. ArchIntern Med 172(5):440–442.

5. Sermondade N, et al. (2013) BMI in relation to sperm count: An updated systematicreview and collaborative meta-analysis. Hum Reprod Update 19(3):221–231.

6. Garg A (2004) Acquired and inherited lipodystrophies. N Engl J Med 350(12):1220–1234.

7. Agarwal AK, et al. (2002) AGPAT2 is mutated in congenital generalized lipodystrophylinked to chromosome 9q34. Nat Genet 31(1):21–23.

8. Agarwal AK, et al. (2011) Human 1-acylglycerol-3-phosphate O-acyltransferase iso-forms 1 and 2: Biochemical characterization and inability to rescue hepatic steatosis inAgpat2(-/-) gene lipodystrophic mice. J Biol Chem 286(43):37676–37691.

9. Magré J, et al.; BSCL Working Group (2001) Identification of the gene altered inBerardinelli-Seip congenital lipodystrophy on chromosome 11q13. Nat Genet 28(4):365–370.

10. Ito D, Suzuki N (2009) Seipinopathy: A novel endoplasmic reticulum stress-associateddisease. Brain 132(Pt 1):8–15.

11. Windpassinger C, et al. (2004) Heterozygous missense mutations in BSCL2 are associatedwith distal hereditary motor neuropathy and Silver syndrome. Nat Genet 36(3):271–276.

12. Ito D, Fujisawa T, Iida H, Suzuki N (2008) Characterization of seipin/BSCL2, a proteinassociated with spastic paraplegia 17. Neurobiol Dis 31(2):266–277.

13. Chen W, et al. (2009) The human lipodystrophy gene product Berardinelli-Seip con-genital lipodystrophy 2/seipin plays a key role in adipocyte differentiation. Endocri-nology 150(10):4552–4561.

14. Liu L, et al. (2014) Adipose-specific knockout of Seipin/Bscl2 results in progressivelipodystrophy. Diabetes, 10.2337/db13-0729.

15. Cui X, et al. (2011) Seipin ablation in mice results in severe generalized lipodystrophy.Hum Mol Genet 20(15):3022–3030.

16. Cooper TG, et al. (2010) World Health Organization reference values for human se-men characteristics. Hum Reprod Update 16(3):231–245.

17. Lundin C, et al. (2006) Membrane topology of the human seipin protein. FEBS Lett580(9):2281–2284.

18. Shimomura I, et al. (1998) Insulin resistance and diabetes mellitus in transgenic miceexpressing nuclear SREBP-1c in adipose tissue: Model for congenital generalized lip-odystrophy. Genes Dev 12(20):3182–3194.

19. Moitra J, et al. (1998) Life without white fat: A transgenic mouse. Genes Dev 12(20):3168–3181.

20. Koutnikova H, et al. (2003) Compensation by the muscle limits the metabolic con-sequences of lipodystrophy in PPAR gamma hypomorphic mice. Proc Natl Acad SciUSA 100(24):14457–14462.

21. Reue K, Xu P, Wang XP, Slavin BG (2000) Adipose tissue deficiency, glucose in-tolerance, and increased atherosclerosis result from mutation in the mouse fatty liverdystrophy (fld) gene. J Lipid Res 41(7):1067–1076.

22. Péterfy M, Phan J, Xu P, Reue K (2001) Lipodystrophy in the fld mouse results frommutation of a new gene encoding a nuclear protein, lipin. Nat Genet 27(1):121–124.

23. Vallet-Erdtmann V, et al. (2004) The testicular form of hormone-sensitive lipaseHSLtes confers rescue of male infertility in HSL-deficient mice. J Biol Chem 279(41):42875–42880.

24. Wang SP, et al. (2004) Expression of human hormone-sensitive lipase (HSL) in post-meiotic germ cells confers normal fertility to HSL-deficient mice. Endocrinology145(12):5688–5693.

25. Sprando RL, Russell LD (1987) Comparative study of cytoplasmic elimination in sper-matids of selected mammalian species. Am J Anat 178(1):72–80.

26. Oresti GM, et al. (2010) Differentiation-related changes in lipid classes with long-chain and very long-chain polyenoic fatty acids in rat spermatogenic cells. J Lipid Res51(10):2909–2921.

27. Fei W, et al. (2011) A role for phosphatidic acid in the formation of “supersized” lipiddroplets. PLoS Genet 7(7):e1002201.

28. Wolinski H, Kolb D, Hermann S, Koning RI, Kohlwein SD (2011) A role for seipin inlipid droplet dynamics and inheritance in yeast. J Cell Sci 124(Pt 22):3894–3904.

29. Tian Y, et al. (2011) Tissue-autonomous function of Drosophila seipin in preventingectopic lipid droplet formation. PLoS Genet 7(4):e1001364.

30. Fei W, Du X, Yang H (2011) Seipin, adipogenesis and lipid droplets. Trends EndocrinolMetab 22(6):204–210.

31. Liu M, et al. (2012) Transient scrotal hyperthermia induces lipid droplet accumulationand reveals a different ADFP expression pattern between the testes and liver in mice.PLoS ONE 7(10):e45694.

32. Chen Y, et al. (2009) Functions of TAM RTKs in regulating spermatogenesis and malefertility in mice. Reproduction 138(4):655–666.

33. Han X, et al. (2007) Alterations in myocardial cardiolipin content and compositionoccur at the very earliest stages of diabetes: A shotgun lipidomics study. Biochemistry46(21):6417–6428.

34. Yang K, Cheng H, Gross RW, Han X (2009) Automated lipid identification andquantification by multidimensional mass spectrometry-based shotgun lipidomics.Anal Chem 81(11):4356–4368.

Jiang et al. PNAS | May 13, 2014 | vol. 111 | no. 19 | 7059

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Mar

ch 3

1, 2

020